Translating a Radiolabeled Imaging Agent to the Clinic

Abstract

Molecular Imaging is entering the most fruitful, exciting period in its history with many new agents under development, and several reaching the clinic in recent years. While it is unusual for just one laboratory to take an agent from initial discovery through to full clinical approval the steps along the way are important to understand for all interested participants even if one is not involved in the entire process. Here, we provide an overview of these processes beginning at discovery and preclinical validation of a new molecular imaging agent and using as an exemplar a low molecular weight disease-specific targeted positron emission tomography (PET) agent. Compared to standard drug development requirements, molecular imaging agents may benefit from a regulatory standpoint from their low mass administered doses, they nonetheless still need to go through a series of well-defined steps before they can be considered for Phase 1 human testing. After outlining the discovery and preclinical validation approaches, we will also discuss the nuances of Phase 1, Phase 2 and Phase 3 studies that may culminate in an FDA general use approval. Finally, some post-approval aspects of novel molecular imaging agents are considered.

Graphical Abstract

1.0. Introduction

The hope behind most preclinical imaging projects is that one day the agent under development will be translated into the clinic for the benefit of patients. That road from initial discovery and validation to clinical use is long, winding, and expensive. Often researchers are discouraged from even trying while many groups simply do not have the broad resources that are required to make the translation a reality. However, the satisfaction of seeing an imaging agent being used for patient benefit is immense and well worth the efforts involved. It is in this optimistic spirit that we draw on our collective experience and describe the steps needed to translate an imaging agent to the clinic. While there are many valid types of drugs and imaging agents that can be considered for clinical translation, including radionuclide-labeled antibodies or nanoparticles, magnetic resonance (MR) and computed tomography (CT) contrast agents, fluorescently labeled agents and ultrasound microbubble agents; of necessity, we must narrow the topic covered by this review to low molecular weight, specifically disease-targeted, radiolabeled molecular imaging agents, for the sake of a coherent story. However, it will be apparent that many of the preclinical and translational aspects discussed are quite similar across many different types of agents.

We will focus on radiolabeled agents, specifically small molecules that target cancers and that fall under the purview of the US FDA regulations. This topic is highlighted because there is a critical need for such agents that address specific disease management issues. Two recent and successful examples of such agents, ⁶⁸Ga-DOTATATE (targeting neuroendocrine tumors) and ¹⁸F-DCFPyL (that targets prostate cancers) illustrate the importance and the impact such agents can have. It is fair to say in both cases these PET imaging agents have transformed the cancers they are targeting by providing clinicians with a far more accurate assessment of the extent of disease than was possible with imaging options that existed before their entry. Importantly, both these agents have therapeutic partners, [¹⁷⁷Lu]-DOTATATE (Figure 1) and [¹⁷⁷Lu]-PSMA-617 (Figure 2), that have improved overall survival and quality of life in patients with neuroendocrine tumors and metastatic castration resistant prostate cancers, respectively [1,2]. This diagnostic-therapeutic partnering serves as a general model for the future (theranostics) even though the approval pathways for a diagnostic and therapeutic are and will remain somewhat different.

Figure 1.

The structure of [177Lu]-DOTATATE, showing the 177Lu radionuclide bound by the macrocyclic chelating agent DOTA on the N-terminus of an octreotide peptide analog of somatostatin. Note the comparison with Figure 8 regarding the different chelates and radiometals, the tyrosine [here] for phenylalanine, and the threonine [here] for threitol, peptide sequence differences.

Figure 2.

Structure of the PSMA-617 targeting agent bearing the macrocyclic chelate DOTA for radiolabeling with ⁶⁸Ga and ¹⁷⁷Lu.

1.1. Beginning in the end

The most important aspect of translating an imaging agent to the clinic is selecting an agent that addresses an important clinical question that will influence how the patient and their physician thinks about and treats the disease. Because the process to the clinic is quite onerous only the most impactful of imaging agents will be translated. Agents that are exceptional in showing an important, but minor feature of a disease and therefore, do not provide useful information to a relevant number of patients, are doomed from the start. Not all imaging agents studied should be translated. In developing a new imaging agent, it is also important to understand the current state of the target disease, recognizing that this is likely to change over the lengthy time required to translate a novel imaging agent to the clinic and that it is possible that either thinking about the disease will change or discovery of a superior agent will supplant an agent in development. There is a certain amount of risk that must be accepted in any such project and every effort should be made to anticipate and minimize the risk involved from the onset.

A superior agent is one that provides new information by revealing more extensive disease with higher sensitivity than conventional imaging. But practically, any new agent must also minimally disrupt the current clinical workflow. For instance, an agent that requires a preparatory injection the day before the actual injection will face resistance because of its impact on workflow. Similarly, an agent that requires several days of uptake before it can be imaged will flunk the “parking lot test”, i.e., patients will tolerate only a limited number of visits to a health care provider for one test. This is particularly true for a diagnostic agent as opposed to a treatment. Therefore, an agent administered on the same day as the imaging with uncomplicated dosing is most likely to prevail. This requirement often necessitates the use of small molecules because of their rapid uptake and background clearance while militating against the use of antibodies and larger antibody fragments or nanoparticles that may take days to target and clear [3,4]. Naturally, there are exceptions to this general preference, for instance if a larger molecule provides some unique information that is impossible to achieve with a smaller molecule.

There are several compelling reasons to develop and use radiolabeled imaging agents. One is that by providing useful biological information on target function that is not found using morphologic data obtained through conventional imaging such as CT or MRI, one can infer more about disease status. This can directly influence clinical decisions that may be tailored to the individual. For example, a tumor with minimal uptake of radiolabeled activity might be monitored closely because the lack of activity may correspond to non-aggressive behavior, while a tumor with robust imaging uptake might prompt immediate radiation or surgical intervention to control growth [5]. In addition, insights on real-world cancer behavior can be gleaned from effective molecular radiotracers. Detecting initial tumor recurrence and early metastatic distribution after curative therapy or assessing response to novel treatments are clinically valuable data that can be acquired through radiolabeled agents.

Tracking a pharmaceutical through the body is another appealing possibility unique to radiolabeled imaging agents [5]. Adding a radioactive element to a drug compound or a similar formulation at microdose quantities allows visualization of the drug’s metabolic route and history while inside the body. Understanding a drug’s dynamic path not only helps confirm proper targeting but can also be used to evaluate the likelihood of success in treating disease. Furthermore, this information can be used to adjust dose regimens, as well as recognize organs at risk for toxicity [6]. This can also benefit early drug development where it can be used to test proof-of-principle and enable the pursuit of more promising drugs while abandoning less promising products. Moreover, radionuclide therapy, where highly potent particle-emitting radionuclides are employed for targeted radiotherapy, often depends on molecular imaging to guide its use. The desire to optimize the power of diagnostic radiotracers and combine them with radioactive therapeutics is developing rapidly and has subsequently evolved into its own field known as Theranostics [7,8].

Ultimately, market size is important to consider at the beginning of imaging agent development. Given the expense of developing a new agent, there must be a reasonable expectation of return on investment, which is typically governed by the number of affected patients who may benefit from the test and the frequency the test will be needed. Therefore, the more common cancers are clearly preferred targets for the development of imaging agents. Those with a long natural history will require more scans. Secondary uses in less common cancers can also contribute to the return on investment of an imaging agent. More specifically, from a commercial perspective, the agent should have broad application against most sub-types of a particular cancer since a more restricted target or indication limits potential application from the outset. Markets are also sensitive to treatment paradigm changes, and the introduction of effective radiopharmaceutical treatments can increase adoption of diagnostic imaging that informs the therapeutic agents. A dual approval track, often referred to as a companion diagnostic, is likely to become more prevalent as the age of molecular medicine continues to develop.

Finally, it is difficult to successfully implement any change in medicine, let alone a diagnostic imaging intervention, without the support of clinicians who become the standard bearers for the agent and its use. These clinicians take care of the patients with the diseases in question and are passionate about improving care. They will generate interest in the agent at meetings and in the literature while it is still in development so that by the time it becomes commercially available there will be a pent-up demand for it. It is vital that such champions be engaged early in the process, not just for this reason, but to also ensure that the imaging agent is addressing a critical need of the end-user: practitioners and patients.

When developing an agent for cancer imaging, maximal sensitivity and specificity are clearly desired but are rarely possible in one agent. In cancer, false negatives are a way of life and are well understood by oncologists as a limitation of any technology. In contrast, false positives are poorly tolerated as they can necessitate additional procedures such as biopsies that may be psychologically stressful to the patient. Therefore, every effort should be made to design agents that have higher specificity even at the expense of sensitivity. Interestingly, this practical truism is counter to the theoretical consideration where one might think that false negatives represent the greater problem, in that they imply the missing of disease sites and thereby underestimating disease extent and seriousness.

1.2. Discovery

There are a marvelous and bewildering number of ways to identify potential targets for useful imaging. Most targets are in the cell membrane where they are accessible via intravenous injection. It is possible to develop cytoplasmic and even nuclear targets, but these entail more complexity. Discovery sources of potential targets include immunohistochemistry, functional studies, genomics, transcriptomics, proteomics, metabolic studies, etc. Among these, immunohistochemistry provides the best overview of expression and distribution of the target. Practically, any membrane target, whether functional or nonfunctional is a potential option for an imaging agent. Expression levels and consistency across different tumors of the same histology and across the same tumor over time are important considerations and again, immunohistochemistry often proves the decisive tool. On occasion, a target is meaningful because it indicates potential sensitivity to a particular drug. The target should be broadly expressed across many stages of the cancer’s natural history and its expression should be high enough to be a reliable indicator of disease. A full discussion about how to identify suitable ligands for targeting cancer cells is well beyond the scope of this review and can be accessed elsewhere [9,10]. Meanwhile, any chosen target should be specific for the disease entity and ideally should not be broadly expressed either in normal tissues or in inflammatory conditions [11].

However, it must be admitted that there is probably no such thing as a wholly cancer-specific antigen target, as most cancers induce up-regulation of disease-associated antigens as compared to the same antigens on normal tissues. In the opposite instances of antigenic down-regulation, it is difficult to envision an imaging agent that could identify such a subtle expression change. Interestingly, this ‘cancer-specificity’ paradigm has been relaxed in recent imaging agent research as investigators have also focused on cells that might support and/or enable cancer growth, such as fibroblasts and immune cells such as B and T lymphocytes [12,13].

1.3. Radiolabeling and Platform Technologies

There are a variety of newer methods to develop novel molecular moieties capable of binding any designated target as the field has expanded exponentially over the last few decades. Originally and historically, imaging agents for most cancers were based on the grosser physical properties of these cancers, for instance size and blood flow, to detect their presence [14-16]. The advent of more sophisticated molecular targeting of cancers for diagnostics, generally, might be traced to success in identifying changes in brain and neural chemistries. In those applications researchers found that they had access to highly specific individual agents targeted to specific systems, such as acetylcholine, dopamine, and serotonin [17-21]. Hence, target specificity is built into the basic biochemistry of the models for such indications, although this is likely to remain the major application for carbon-11 [¹¹C]-labeled agents going forward due to that nuclide’s 20-minute half-life. Because such agents target tiny amounts of receptors, carrier-free radiolabeled tags (i.e., radionuclides free from stable isotopes) were the ideal tracers and it was imperative that they be generated with no or little non-radiolabeled compounds present. Much of this neurological work was carried out with ¹¹C- and fluorine-18[¹⁸F]-radiolabeled analogs of the native compounds, since in the former instance, exact ¹¹C analogs of the carbon-12 natural compound could be prepared while in the latter instance replacement of an organic compound’s single hydrogen atom with a single ¹⁸F atom represented the smallest possible structural alteration and resulted in compounds widely believed to generally behave in the same manner as the original compound. In fact, this assumption was found to be not always true as in this example, [¹¹C]-choline is mostly excreted by the pancreas whereas [¹⁸F]-choline is mostly excreted by the kidney [22]. Fluorine replacing hydrogen in any molecule may represent on paper a minimal atomic change to that molecule, but it is still catabolized differently, most notably often undergoing defluorination reactions leading to free ¹⁸F, which itself is a bone-seeking agent. This pharmacokinetic and biodistribution aspect must always be borne in mind when considering [¹⁸F]-radiolabeled imaging agents.

As new agents outside the realm of neuroscience were considered both the short 20-minute half-life of ¹¹C, along with the complex and limited organic chemistry needed to radiolabel many organic molecules rendered it impractical for more general use. The longer lived (109.8-minute half-life) ¹⁸F was more practical although the highly complex chemistry often needed for radiofluorination precluded wider early application until better radiochemistry methodologies were developed. The half-life timeframe for imaging radionuclides may be extendable to as much as 10 days, but shorter half-lives are certainly helpful to reduce patient radiation exposure. The chemistries for ¹⁸F-radiolabeling methodologies have developed tremendously over the past 30+ years and ¹⁸F radiolabeling is now much more practicable in terms of reproducible radiolabeling yields, and achievable within useful timeframes given the 109-minute radionuclide half-life involved. At the chemically most basic level, reactive fluorine might be added to organic entities via either electrophilic (F⁺) or nucleophilic (F⁻) chemical reactions and while the former was investigated in the early days it has been almost completely superseded by the nucleophilic ¹⁸F⁻ chemistry for the simple reason that the electrophilic ¹⁸F⁺ ion is obtained from F-18 fluorine gas, which is extremely dangerous and difficult to handle (e.g., fluorine gas dissolves glassware). What prevented faster use of nucleophilic ¹⁸F⁻ chemistry was the profound non-reactivity of the ¹⁸F⁻ anion, particularly in the presence of water. Although fluorine is considered the most electronegative element, the fluoride anion in water is shielded by a ring of hydronium ions, water molecules and counter-ions rendering it chemically inert. Until these reactivity issues were addressed adequately, progress toward radiofluorinated molecules was halting and slow [23].

As an aside, what may seem like obvious alternatives to a radiofluoride solution, utilizing radioiodination [24] or radiobromination [25,26] turns out not to provide a practical answer. Production of radioiodinated products requires special facilities with highly reliable venting. Personnel need to be monitored for inadvertent exposure to volatile iodine. Radiobromine is difficult and costly to make. Finally, the half- life and energy emissions of the various iodine and bromine radioisotopes are poorly suited matches for optimal clinical imaging.

2.0. Chemical Radiosynthesis and Initial Testing

Having identified a disease target suitable for a potential imaging agent, it is important to label that agent in a manner that does not impede or degrade its initial affinity for the target. Presently, radiochemists are skilled and experienced at understanding where it is possible to radiolabel a molecule without interfering significantly with its function. The radiolabeling itself should or must be straightforward and potentially automatable for clinical development and general, practical application. Chemically, radiolabeling techniques can be split into two main types, covalent chemical reactions particularly using ¹⁸F, and complexation reactions applicable to radioactive metal ions. With > 70% of the elements represented by metals it follows that radiometals are the larger group of radioactive elements for the compounding of radiolabeled targeting vectors. A brief list of suitable imaging radionuclide elements of potential use with molecular targeting agents from over 1200 known radionuclidic elements is listed in Table 1. In the early days many agents were developed for single photon emission computed tomography (SPECT), but that imaging approach is being superseded by positron emission tomography (PET) due to the higher sensitivity, precision, and quantification of the latter imaging technique. As such, it is likely that the SPECT-based nuclide agents, like the F⁺-18 radiolabeling approach, will become more of historical interest going forward. Over the last 20 years, the dramatic developments in PET imaging systems [27-29] and parallel discovery of the tremendous utility of the radiofluorine analog of D-glucose (6-[¹⁸F]-fluorodeoxyglucose) for imaging glucose metabolism [higher in tumors] have powered the drive to further develop novel PET agents, particularly based on ¹⁸F [30].

Table 1.

Selected radionuclides used in molecular imaging research both for SPECT and PET imaging. Metal cations require a bifunctional chelating to be bound to targeting vectors, enabling a stable radionuclide complex to be formed, with the nature of the chelating agent dependent on each individual metal ion.

Nuclide	Half-Life*	Emission(s) (keV)	Chemical Form	Valence
F-18	109.8 m	511 (PET)	Fluoride	−1
Cu-64	12.7 h	511 (PET) + others	Metal ion	+2
Cu-67	62 h	185(SPECT) + others	Metal ion	+2
Ga-67	78.3 h	184(SPECT) + others	Metal ion	+3
Ga-68	68 m	511 (PET)	Metal ion	+3
Tc-99m	6 h	140 (SPECT)	Metal ion	0 to +7
In-111	67.2 h	111(SPECT) + others	Metal ion	+3
I-123	13.2 h	159 (SPECT)	Iodide	−1
I-131	8 d	606 (SPECT) + others	Iodide	−1
Zr-89	78.4 h	511 (PET) + others	Metal ion	+4

^*m = minutes, h = hours, d = days.

2.1. Radiofluorination Chemistry

From imaging quality and dosimetry viewpoints F-18 is the preferred radionuclide since it is a pure 511 keV positron emitter while other potential PET radionuclides also have attendant high energy gamma energy emissions that can degrade image quality and increase attendant patient doses, and/or particulate emissions that can also imply the latter effect. As mentioned, the basic problem with radiofluorination is the profound poor reactivity of the fluoride anion toward nucleophilic substitution of activated organic precursors in water. A typical precursor that does work is exemplified by the Balz-Schiemann reaction [31] where a highly reactive aromatic diazonium salt is replaced by ¹⁸F-fluoride with the formation of nitrogen gas (N₂) as the ultra-stable side-product and providing an irreversible forward reaction driving force. That is, extremely reactive leaving groups as exemplified by diazonium salts were necessitated by the fluoride ion’s low reactivity as a nucleophile and this strategy was paired with extensive efforts to keep radiofluoride reaction mixtures completely dry [32]. Historically, some key achievements and advances in improved radiofluorination chemistries may be highlighted: In a landmark study in 1984, a major pioneer in the field, Michael Welch¹¹, [17] and coworkers reported syntheses of no-carrier added haloperidol derivatives using triazene leaving groups. Key methodology was that ¹⁸F obtained from cyclotron production in aqueous solutions was dried several times using azeotropic distillations to dryness and reconstitution into organic solvents favorable to fluorination reactions such as alcohols, acetonitrile and dimethylformamide (DMF). It must be remembered that when performing these very ‘hands-on’ tasks pioneering scientists were needfully working very quickly with very high radioactive doses of fluorine-18 since natural decay inevitably reduced the activity amounts obtained in any final product and that the 511keV positron emission was very difficult to shield adequately, even with very thick lead [33,34]. In one of the most useful insights in the history of radiofluoride chemistry, Hamacher and colleagues [35] synthesized and used [¹⁸F]-FDG by applying the catalytic additive Kryptofix 222, the cryptand dramatically increasing radiofluorination yields, and used it along with a triflate leaving group on a mannose precursor to the glucose product, with inversion of configuration at the reactive carbon atom, generating the radiolabeled glucose (Figure 3). Most notably, the agent Kryptofix 222 acted as a cryptand surrounding the metallic cation and separating it from the ¹⁸F⁻ reactant, thereby maintaining the latter’s nucleophilic reactivity by protecting it from deactivating hydronium/hydroxide ions that may be present even in extremely dried and re-dried solutions. Vaidyanathan and Zalutsky [36] in 1992 described preparation of an N-hydroxysuccinimide ester precursor of an [¹⁸F]-fluorobenzoate useful as an intermediate hapten for easy further conjugation to targeting agents. This synthetic strategy change enabled production of an intermediate pre-activated active ester recognizable to standardly trained organic chemists, that already had the radioactive atom attached, and just requiring linkage of this intermediate to a designed targeting agent that had a free amino group for a peptide-like coupling reaction. By insertion of another reaction into the overall schema the generation of this active ester intermediate implied that a longer time, post-radiolabeling, would be needed to obtain any desired radioactive targeting agent, but this may be offset by the advantage of performing a well-known standard peptide-type coupling reaction as a final step. The safety handing issues and the rapid 109-minute decay of the ¹⁸F must constantly drive the developmental chemist to perform coupling reactions between the preformed radiolabeled hapten and a selected targeting vector in as rapid and high yield as achievable. In a further significant advance to the basic approach, Cuthbertson et al. in 2010 described extension of the prelabeled hapten approach to more water-friendly radiofluorinated pyridinyl tetrafluorophenyl active esters based on nicotinic acids, resulting in higher yields, faster reaction times and stable products, and this approach has become a standard methodology in the field [37].

Figure 3.

Landmark reaction to generate radiofluorinated [¹⁸F]FDG by radiofluoride SN2 replacement of the triflate leaving group on a protected mannose precursor to generate the fluorinated glucose with inversion of configuration at the reaction center. Aside from the very active triflate leaving group the reaction is optimized by initial drying of the [¹⁸F]-fluoride into acetonitrile and addition of the cryptand Kryptofix-2.2.2 to protect the fluoride from reaction quenching hydronium species.

More recently, a slew of newer methods effecting for radiofluorinations of targeting agents both directly [38] and indirectly have been described. These methodologies are very eclectic and include the radiofluorination of multiple intermediate haptens such as aluminum complexes [39,40] radiofluorination reactions of boron derivatives [41,42] and of silica derivatives [43,44] prior to attachment of said intermediate complexes to targeting vectors. In addition, various other technologies such as electrochemistry [45], photochemistry [46] and microfluidic chemistry [47] are being actively investigated as technologies applicable for the improved rapid radiofluorination of promising imaging agents. It’s too early to discern whether any of these approaches will become standard with new agents.

Once a basic radiofluorine chemistry is established as robust and repeatable, as with [¹⁸F]-FDG or any other radiofluorinated targeting vector, that chemistry technology is suitable for research purposes, but significant additional tasks are required to make it suitable for practical application up and toward clinical adoption. Firstly, scale-up to Curie level production and/or employment within automated systems for production of the radiofluorinated agent under sterile and pyrogen-free conditions is needed. Originally, CTI (Knoxville, TN, now Siemens PETNet) developed a euphemistically called ‘black-box’ comprised of various lead safety-shielded vials, columns and tubing arrangements designed to mix precursors with ¹⁸F- and other required reactants for the processes under computerized control. The entire systems had to generate a purified radiofluorinated targeting agent meeting pharmacopeia and regulatory requirements such as purity, radiopurity, sterility and apyrogenicity. As such, the production processes were properly shielded and designed to reduce user errors while incorporating all necessary steps to generate a desired clinically usable product. At the present time, about a dozen such synthesis boxes exist (e.g., GE TRACER lab, TRASIS, Synthera, IBA, SOFIE, et.) to provide various radiolabeled targeting vectors with each company’s ‘black-box’ arrangement often designed specifically for each individual agent’s reaction, purification, and quality control requirements. A typical system is shown in Figure 4. Within a general radiofluorination box, components are often supplied as single use, GMP-compliant cassette components, for instance single use vials, tubing, connectors, and syringes which are often linked together and shipped to radiopharmacies as ‘plug-and-play’ plastic-wrapped fully sterilized components.

Figure 4.

A typical ‘black box’ radiosynthesis set-up for generating radiolabeled targeting agents by remote, automated, computer-mastered systems. This is an example of the General Electric TracerLab system. Briefly, the inside of the system is a series of vials, columns, tubing, heaters, etc, to enable mixing of the reaction components and their later purification into the final product. The choice of the actual system components is determined by the specific chemistries needed to furnish the desired final product. Reprinted from Nature Protocols, 2019.

As mentioned above, it has been found that a very important aspect of radiofluorinated targeting agents is the stability of the radiofluorine-carbon bond, since cleaved radiofluoride is rapidly and specifically taken up in cortical bone, hence the original application of ¹⁸F-18 fluoride ion itself for bone imaging [48]. Since many primary cancers metastasize to bone, identification of metastases in the presence of a less-than completely stable radiofluorinated targeting agent presents a significant problem. That is perhaps, a minor problem to publishable optimized research, but it is also a more serious, perhaps lethal defect to any thoughts of a generally applicable clinically and commercially viable agent. From early and continuing work, it is now more fully appreciated that at the basic chemical level ¹⁸F- probably needs to be attached to an aromatic ring on or within the targeting agent [49], as aliphatic radiofluoride attachments have often led to pronounced in vivo defluorination and bone uptake although, as cited above, current active research is methodologically diverse. Certainly, other aspects of any prospective targeting agent including its own structure and function may also lead to unwanted bone deposition and these must be evaluated individually for each contemplated reagent, but the basic radiofluorine radiobiological chemistry discussed above is always a factor.

2.2. Radiometal Radiolabeling Options

Whereas development of useful radiofluorinations have required extensive, specific, and specialized expertise in fluorine radiochemistry, radiometal labeling may often be performed much more simply by mixing a radiometal ion with a strong chelating agent that is coupled to a targeting agent (or ready to be coupled to a targeting agent). This is preferably performed as a ‘last-step’ radiolabeling, after the chelate is functionalized for covalent coupling to a targeting agent and coupled thereto with retention of targeting agent viability. However, prior to application even at a research level, radiometal-chelate-complex stability must be established under stringent in vitro and In vivo challenge conditions. In other words, there is significant chemistry, and metal biology expertise, and knowledge lying behind what will finally become a simple, one-step radiolabeling protocol using an added radiometal ion. A full discussion of all aspects of radiometal radiolabeling of multiple targeting vectors is beyond the scope of this review [50,51] and only the most pertinent practical examples for clinical development will be focused on below.

At this point in the current development of radiometal-tagged targeting agents, the emphasis is mainly directed toward PET imaging using the [⁶⁸Ga] and [⁸⁹Zr] radionuclides as being of most clinical interest. Interestingly, the two elements are quite different chemically but chelates for both have grown from their original relationship to naturally found ferric ion, since gallium is almost the same ionic size as ferric ion and [⁶⁷Ga] as a SPECT candidate radionuclide, most basically injected as an iron analog [52], preceded by some decades the current interest in the positron emitting [⁶⁸Ga]. The iron-chelating agent deferoxamine (Desferal^®) (Figure 5) was originally developed in the 1950s for administration to patients suffering from Cooley’s anemia, a condition involving a genetic inability to produce and utilize hemoglobin [53]. This meant that patients were treated with regular infusions of whole blood which often left them suffering from iron overload, as the body has no established way of excreting large amounts of iron and chelation therapy with deferoxamine became the gold standard of treatment [54]. In fact, this type of therapy involving administration of chelating agents to address metal intoxication issues also led to development of polyaminocarboxylate chelates such as EDTA, DTPA etc. for detoxification of workers and others who may have ingested radioactive heavy metals, before their adoption for radiolabeling using SPECT emitters such as ⁶⁷Ga and ¹¹¹In [55].

Figure 5.

The structure of the iron-chelating compound deferoxamine, later adapted for use for the chelation of +3 and +4 valence heavy metals such as ^67/68Ga and ⁸⁹Zr. The presence of the free amino group was convenient for chemists who wanted to couple the chelating agent to a targeting vector while the three hyroxamate metal binding centers did not interfere with conjugation reactions under the right conditions and provided a strong hexavalent chelate for 3+ gallium.

Generally, for PET radiodetection purposes deferoxamine is conjugated to various targeting vectors and then radiolabeled with [⁶⁸Ga] by simple mixing of the deferoxamine-vector conjugate with the [⁶⁸Ga] ion in largely neutral and dilute aqueous solutions. Again interestingly, and despite an extensive history, deferoxamine use for [⁶⁸Ga] has been somewhat superseded using polyaminocarboxylate macrocyclic chelates based on the ring systems DOTA and NOTA [56] (Figure 6). [⁶⁸Ga]-NOTA complexes are more stable than [⁶⁸Ga]-DOTA complexes but not so much so that the 68-minute half-life [⁶⁸Ga] cannot be usefully used with a DOTA analog [57]. In addition, use of a DOTA chelate in targeting vectors further allows use for the same chelate for pairing with either [¹⁷⁷Lu] or [¹²⁵Ac] complexes for radiotherapy purposes, which as a theranostic pairing, significantly expands the utility and interest of any novel PET imaging tracer [58-60].

Figure 6.

Structures of the macrocylic ligands NOTA (left) and DOTA (right). Although NOTA is a better size fit for the small ⁶⁸Ga cation and forms more stable complexes than the larger 12-atom DOTA ring the latter can also readily incorporate therapeutic radionuclides such as ¹⁷⁷Lu and ²²⁵Ac. Thus, targeting vectors made using the DOTA chelate are suitable as matched pair theranostics, while the slightly less stable complexes of DOTA agents with ⁶⁸Ga is ameliorated by the short half-life of the radionuclide, rendering them sufficiently stable for PET use given the theranostic advantages.

Zirconium-89 is cyclotron produced and has a usefully long half-life when used with long circulating targeting agents such as monoclonal antibodies [61]. Despite its normal +4 valence it binds reasonably well to deferoxamine-substituted vectors designed for +3 ferric iron and +3 [⁶⁸Ga]. In either instance, from a radiopharmaceutical viewpoint, the chelate-vector complex needs to bind added radiometal, at least at the > 90% minimum incorporation level and preferably >95%. Free, unbound radiometal is not allowable in injectable targeting agents as free metals will distribute themselves in high unwanted amounts in various non-target tissues, often in bone like radiofluoride. If near-full incorporation is achieved the remaining <10% non-bound metal can be sequestered by addition of a low molecular weight free chelate to scavenge unbound radiometal and ensure that this portion of radiometal is quickly flushed through the renal system. As an alternative to sequestration as a ‘purification’ step, various applicable chromatographic methods might also be considered, type-dependent on the nature of the molecule under question although further technical complexity is inherent in such a choice. Of course, parameters for executing a [⁶⁸Ga] versus [⁸⁹Zr] radiolabeling protocols are significantly affected by the relatively short and long half-life values for the two radionuclides.

From a product manufacturing standpoint, targeting vectors, pre-radiolabeling, may be placed in sterile, pyrogen free, sealed vials with excipients and stored at refrigerator temperatures or optionally lyophilized for long-term storage stability. Radiolabeled product is then produced on-site by addition of radionuclide solution and binding of the added radiometal by the strong chelating portion of the complex. As with the radiofluorination reactions above, the chemistry will have been simplified enough at this point that it can be incorporated into remote, computerized set-ups for radiopharmacy use [62-66]. It must always be appreciated by the developing radiochemists that their radiopharmacist end-user colleagues must operate under their own very stringent practice requirements. Any bench chemistry required for radiolabeling is typically met with resistance as a workflow disruption, and likely lies outside the constraints of state and federal radiopharmacy regulatory compliance. Similarly, quality control procedures for purity testing must be reasonable to accomplish in a radiopharmacy setting. Space constraints, excessive capital investment in testing equipment, and specialized training beyond that which is typical for radiopharmacists for quality control procedures may limit the ability for radiopharmacies to offer an agent to patients. Challengingly, all these factors must be considered by the designing chemist/radiochemist in proposing novel agents for potential clinical application.

As opposed to cyclotron-produced zirconium-89, gallium-68 is produced by a radionuclide generator system comprised of the long-lived [⁶⁸Ga] parent, germanium-68, adsorbed onto a solid matrix from which the daughter [⁶⁸Ga] radionuclide is selectively eluted. The first commercially available radionuclide generators were developed for the molybdenum-99-technetium-99m parent-daughter enabling [^99mTc] to become the preeminent and preferred radionuclide for SPECT imaging purposes. In both cases the long-lived parent decays to daughter over a time-period of hours to reach maximum daughter ingrowth prior to elution. Once eluted (‘milked’), the long-lived parent continues to decay to generate more daughter, that can be re-eluted from the generator within several hours of the last elution, or most conveniently each day. These generator systems are so convenient for generating these radionuclides that the generator-derived radionuclides are nearly always available on-site.

A prime example of the ⁶⁸Ge/⁶⁸Ga generator is the regulatory approved system developed over a 10-15-year period by Eckert & Ziegler company. Numerous issues had to be ironed out in generator construction and performance. One particularly notable issue was the demonstrated absence of unwanted, long-lived parent radionuclide ⁶⁸Ge contaminating the ⁶⁸Ga eluate, termed ‘breakthrough’, originally found present when repeatedly using the system and increasing as the generators aged [67]. After considerable background research and reiterations, the generator system has now been modified and streamlined so well that [⁶⁸Ga] generator eluate can be passed directly into the automated radiolabeling boxes where the vector-chelate vials are placed, for remote radiolabeling [68,69]. As an adage, and as with the background chemistry unseen behind targeting agents, a considerable amount of thought and design goes into making clinically useful radionuclide generator systems, and making the processes involved appear simple. There are now multiple samples of radiochelate-based reagents in, or nearing, full and final development, or even clinically approved for use.

2.3. Agents Approved and Under Development

The FDA first approved ¹⁸F- radiofluoride ion for bone imaging back in 1972 when PET and supporting sciences were in their infancies. In 1999 the FDA approved [¹⁸F]-FDG for brain imaging, glucose being taken up in brain very quickly in large amounts, and later extended the label to PET oncologic imaging for “abnormal glucose metabolism to assist in the evaluation of malignancy in patients with known or suspected abnormalities found by other testing modalities, or in patients with an existing diagnosis of cancer.” Despite its widespread study in leading Universities, Hospitals and Research Institutes CMS (Centers for Medicare and Medicaid Services) reluctance regarding recommending cost reimbursements for new diagnostic tests, thereby allowing broader routine use, have been harder to come by. In the US system this is clearly an important factor for this agent and more generally for the development of PET imaging agents. Nevertheless [¹⁸F]-FDG has shown its utility for oncologic PET imaging in thousands of studies and its utility as a contrast agent with a bright future has spurred the research and development of other agents.

In May 2021 the FDA approved the use of [¹⁸F]-DCFPyL (Plyarify) (Figure 7) for suspected metastatic or recurrent prostate cancer with PET, validating an agent first described over a decade ago [70] and showing outstanding PET imaging results at multiple institutions since then [71-73]. The agent targets prostate specific membrane antigen (PSMA) which is upregulated on many prostate cancers and differs definitively by its targeting of this specific tumor-associated antigen from [¹⁸F]-FDG, which targets a metabolic occurrence, namely glucose metabolism. Another long road to final regulatory validation was undertaken for the approval of Octreoscan (Indium-111-pentetreotide) (Figure 8) for the SPECT imaging of neuroendocrine tumors by targeting of the somatostatin (a 14-amino acid cyclic peptide) receptor upregulated specifically on these diverse tumor types [74,75]. From this agent, Gallium-68 DOTATATE emerged which has become the standard imaging agent for neuroendocrine tumors (Figures 9,10) Notably, the imaging studies with the somatostatin analogs have recently led to approval of the related radiotherapy agent LUTATHERA^® (lutetium Lu-177 dotatate) [76] (Figure 1), a somatostatin-targeting peptide, modified for stability and bearing the DOTA chelate for binding the radiometal ¹⁷⁷Lu, and representing a system where the diagnostic/therapy theranostic pair can make the journey through the entire developmental and regulatory processes [77].

Figure 7.

Structure of [¹⁸F]-DCFPyL targeting the PSMA antigen on prostate cancer cells. The agent consists of the [¹⁸F]-nicotinyl hapten attached to the carboxypeptidase G2 enzyme (PSMA)-binding subunit, comprised of a C-terminal glutamic acid, a urea moiety, and a lysine residue.

Figure 8.

[111In]-octreotide showing the [111In] cation bound tightly by the open-chain DTPA chelating agent and attached to the N-terminus of the octreotide peptide targeting vector. It is a di-cysteinyl disulfide-linked cyclic octa-peptide based on the naturally occurring cyclic 14-amino acid somatostatin, with a D-tryptophan substitution within the ring and an unnatural threitol residue at the C-terminus.

Figure 9.

⁶⁸Ga-DOTATATE scan of a patient with metastatic neuroendocrine cancer showing multiple metastases in the tissues of the neck, liver and bones. The PET agent targets SSTR2 expression within the tumors which has minimal expression in normal tissues.

Figure 10.

⁶⁸Ga-DOTATATE scan of a patient with recurrent neuroendocrine cancer in the retroperitoneal lymph nodes but also within the vertebral bodies and skull. This patient went on to receive ¹⁷⁷Lu-DOTATATE (Lutathera™) as therapy.

These recent successes are helping to spark increased searches for the next specific disease diagnostic agent and the possible theranostic pairings that might be available (Table 2) [78]. At the forefront of these efforts are studies on finding imaging and therapy analogs targeted to the PSMA protein [79-83]. Many other targets are being pursued under active investigations [84-86].

Table 2.

Selected PET and related theranostic agents that have received regulatory approvals or are nearing regulatory approvals.

Agent	Name	Commercial Name	Target	Approval Date
68Ga-DOTATATE	DOTATATE	NETSPOT™	Neuroendocrine	08/21/19
68Ga-PSMA-11	TLX591-CDX	ILLUMET™	Prostate cancer, PSMA	12/01/20
18F-DCFPyL	F-18-piflufolastat	PYLARIFY®	Prostate cancer, PSMA	05/27/21
177Lu-PSMA-617	Lu 177-dotatate	Lutathera®	Neuroendocrine	01/26/18

2.4. Preclinical Testing of Potential Radiopharmaceuticals

Following successful chemistry and radiolabeling work, in all instances in vitro and in vivo testing are the next steps in research. Depending on the target, a suitable mouse model of cancer is usually needed. Ideally, the model(s) should display a dynamic range of target expression and should include negative controls. Before animal work is undertaken required in vitro testing usually starts with affinity testing in which dissociation constants (K_D) are established for the agent under study. Usually, low nanomolar k_d is needed for successful in vivo imaging with k_d defined by the sum of kinetic and thermodynamic properties of binding summarized by k_on and k_off rates. Low or sub-nanomolar K_Ds, particularly for small molecules with monovalent binding capability and likely displaying rapid pharmacokinetic clearance, may be considered a minimal requirement [87].

The most definitive tests for demonstration of imaging agent promise are the in vivo ones. The most important in vivo test is the biodistribution study and is typically performed in immune deficient mice with implanted human tumor xenografts [88,89]. A small amount of the radiolabeled agent is injected into the animal which is then euthanized at different time points after injection. The organs including blood are individually collected and are accurately counted on a gamma counter set to appropriate energy windows. After correction for radioactive decay, it is possible to accurately quantitate the amount of uptake within both target and normal organs. The value of biodistribution lies in its ability to identify and quantify potential problems like abnormal high uptakes in the kidneys or liver or retention in the blood or bones. Biodistribution studies are also important for establishing appropriate radiation dosimetry for eventual human translation [80]. It must be borne in mind that targeting a murine xenograft in a mouse may represent a defective model system in that the target protein within the human xenograft may be completely absent in normal mouse tissues. Hence if the target graft tissue is human then the mouse model will tend to underestimate the true organ uptake in humans. In the future, a more directed and relevant approach may be to start with evaluations of the sequences of the mouse and human targets using the available databanks [90]. When there is a high degree of homology there is a chance that the biodistribution will more accurately reflect what will be observed in humans. However, in the more likely instance that the amino acid sequence of the mouse is substantially different from the human, as is now usually the case, normal organ uptake is likely to be underestimated in current mouse biodistribution studies.

Ironically, perhaps the least useful of all the preclinical in vivo testing options are the actual imaging studies usually conducted on a microPET/CT or microSPECT/CT imaging cameras. The uptake of the tumor relative to the other organs can usually be readily seen on such imaging. However, because the images themselves can be adjusted manually to maximize the apparent uptake in the target and minimize the apparent off-target uptake, in vivo imaging that is optimized for other purposes can often be misleading and over-promising as to future clinical utility. It must be borne in mind that tumors implanted in mice are usually much larger with respect to the body size of the animal than are typical human tumors, leading to false assumptions about sensitivity [91]. By using tumor models with different amounts of target expression it should be possible to titrate the imaging uptake, with the highest expressing tumor having the highest uptake followed by the next highest expressing tumor, and so on. In this way, it can be shown that the imaging agent uptake is likely to reflect the actual expression of the target. However, it is important to remember that any tumor-bearing animal model may mislead even at the biodistribution level due to idiosyncrasies in the chosen tumor, its growth, blood supply and size, so applicability to human imaging must be interpreted with caution. In addition, as mentioned above, if the human tumor xenograft is in a mouse that does not express the target antigen on its own tissues, overly optimistic relative uptake and tissue retention assumptions may be made that cannot be replicated in a human bearing the target antigen either on tumors or normal tissues. Nonetheless, this preclinical imaging step is absolutely necessary prior to human studies to demonstrate radiotracer behavior in a living system and imaging feasibility on a small scale.

For certain applications it may be desirable to demonstrate that a particular treatment alters the uptake of a test tracer (e.g., reduction of tumor uptake because of successful treatments). For such studies, it is useful to scan the same animal before and after treatment using identical procedures. This will help ensure that the changes seen reflect actual response to the therapy and not artifacts due to other factors (e.g., tumor necrosis). Ultimately, this pairing of a PET imaging agent as an integral part of a molecularly targeted therapeutic agent regimen represents an ideal application of the imaging modality and can be demonstrated preclinically prior to its clinical extension [92].

3.0. Reduction to Practice

Having established that the agent has high affinity, high selectivity, and favorable biodistribution, the next step is to consider optimizing the chemistry to practice, i.e., designing the chemical synthesis in a way that makes it commercially viable. Typically, this means reducing the number of steps involved and improving the synthetic yield. A variety of synthetic options and variations can and should be considered. Reflecting on the likelihood that commercial production will be overseen by technicians, not necessarily chemists, is also important. The synthetic processes must be simple and robust and potentially adaptable to automated synthesis with minimal human intervention. Complicated syntheses will fail frequently and if this happens enough times, the agent will be abandoned by clinicians. Thus, from the outset attention should be paid to the robustness of the radiosynthetic methodologies.

It is also important that the inventors submit an invention report with the intent of patenting the novel product composition and the processes employed in preparing the new agent. Even if the motivation for the agent is not financial on the part of the inventors, it is important for them to secure intellectual property (IP) coverage as failure to do so will make the product commercially unattractive to entities seeking to license and develop the product but requiring exclusivity for protection of their financial investment. At this stage it is important not to disclose the invention publicly until the IP is secure as this can nullify any potential patent. The patent application must include the synthetic and other methods used along with all relevant preclinical data, potential clinical applications, and provide usage examples embodied by the claims to the invention.

3.1. Clinical/Commercial Aspects

Turning to more clinical and commercial aspects, for any radiopharmaceutical to be successful, the raw materials for scalable, commercial manufacture of the drug must be available or producible in sufficient quantities to meet demand. This supply must be reliable and consistent and such issues often condemn as impracticable proposed radionuclides that may have desirable physical properties in themselves. This principle applies to radioactive and nonradioactive components, excipients/inactive ingredients, and even container closure systems. There are many historic examples of significant raw material shortages having a detrimental worldwide impact on nuclear medicine facilities. Such examples include a multimonth shortage from the sole US supplier (Jubilant Radiopharma) of human albumin needed to make macroaggregated albumin for lung perfusion imaging. The most notable raw material shortage of current times occurred during 2008 and 2009. One of the five nuclear reactors supplying the bulk of North American medical grade parent molybdenum-99 for use in ⁹⁹Mo/^99mTc generators, experienced a heavy water leak, and brought much of ^99mTc-based SPECT nuclear medicine to a grinding halt until repairs were safely completed. Since 2009, much progress has been made to address the ⁹⁹Mo supply chain fragility, with the creation of an outage reserve capacity [93]. Noteworthy efforts to bolster supply include bringing additional reactors online for medical use, minimization of down time via coordination between reactors for scheduled maintenance events, as well as FDA-approval of non-reactor sourced Mo99 for a novel 99Mo/99mTc generator system supplied by Radiogenix Corporation.

3.2. The “Valley of Death”

Many molecular imaging agents are proposed and described each year in the scientific literature. A minority of those may pass basic scientific tests of affinity, biodistribution and pharmacokinetics when considering further development. However, only a tiny minority of the proposed and viable candidates ever go forward to clinical testing. This phenomenon has been called the “Valley of Death” because so many promising agents never make it past this point. There are many reasons for this, but one major reason is that funding for such development work is hard to find and the amount of money involved is considerable. It is completely understandable when one grasps that the investment of the early researcher is typically ~$10⁵, whereas commercialization is typically > $10⁸. Between those two investments lies the “Valley of Death” where many viable agents can no longer be developed [94]. Further complicating this general drug development picture is that the potential return on investment for imaging agents is lower than for most drugs because they are used infrequently and at uneven intervals, reducing the number of suitors interested in commercialization.

To advance a molecular imaging agent out of the laboratory and into the clinic requires substantial investment. Practically speaking, the principal components of this investment are Good Manufacturing Practice and at early stages of development, Good Laboratory Practice in the production of the agent followed by toxicity/toxicology studies performed using GLP/GMP agents, and submission of all data to regulatory authorities in an application for Investigational New Drug (IND) status. With an IND for the specific agent in place, clinical Phase 1 and Phase 2 testing can begin. These steps will be discussed in order.

4.0. Good Manufacturing/Laboratory Practice Production

An option in some institutions is to use a Research Drug Research Committee (RDRC), which can be very attractive to simply enable initial testing in humans but also entails a narrower criteria for use (https://www.fda.gov/drugs/science-and-research-drugs/radioactive-drug-research-committee-rdrc-program). The RDRC approach cannot be used for novel compounds undergoing ‘first in human’ trials. Put succinctly, RDRC is a viable short-term solution if one has no idea whether the agent will work in humans. It can provide valuable preliminary data. However, if investigators feel the agent is likely to be successful clinically then a Phase 0/1 approach may be better, and of course, if one hopes to bring the product to the clinical market, an IND will still be needed.

The FDA mandates that any agent that will enter a patient’s body under an IND (outside of RDRC) be manufactured according to current Good Manufacturing Practices (cGMP). For toxicity studies, the major difference between GMP and GLP (good laboratory practice) production is the amount of documentation required, but the chemistry requirements are equally strict. For GMP/GLP production, there are three major requirements: the components should be of high quality, often referred to as “medical grade”, the environment in which the chemistry is performed must be closely monitored for sterility and air purity, typically in clean rooms and the end product is thoroughly tested for stability, sterility and for the absence of endotoxin. Technically, only the last two steps of a synthesis need to be “cGMP” for the product to be considered cGMP. However, in general, the highest quality chemicals should be used throughout to avoid contamination with endotoxin or microbes. The synthesis needs to take place in a certified clean room with its own highly filtered (HEPA) air flow. Typically, a cGMP-compliant clean room consists of a changing room, a sterile intermediate chamber, and the actual clean room where the chemistry is conducted. Typically, products generated in a clean room are transferred to preparation rooms using pass throughs in the wall. The chemistry itself is preferentially performed in a hood where the air quality and air turnover are even higher in quality than the air in the clean room itself. Typically, the clean room is rated at ISO7 and the hood at ISO5, which are measures of air purity by particulate concentration levels. The radiochemist in the room must be completely gowned, gloved, and masked to reduce contamination. To perform GMP production, the radiochemist must be trained in GMP technique and documentation. GMP clean rooms are expensive to build and maintain as they require separate air exchangers and constant monitoring. Also expensive is the skilled manpower needed to work in such facilities, including at a minimum a radiochemist and a radiopharmacist, who will assess the product independently before approving human administration. As a result, relatively few institutions have GMP radiochemistry facilities and commercial sites are typically quite costly to utilize.

The end-product, or the test article, in the parlance of drug development, must be thoroughly tested for identity and purity (https://www.fda.gov/media/136480/download). The amount of radioactivity and mass must be assessed. The product is tested for sterility and endotoxin the latter of which can be quickly assessed. Because most radiolabeled compounds have short half-lifes, there is not time for full sterility testing to be accomplished before injection. Therefore, sterility is often determined several days after the injection. In the unusual circumstance that the microbiological test cultures turn positive, the patient can be carefully followed by their physician for signs of infection or placed on prophylactic antibiotics. For drug products whose manufacture is completed entirely at the GMP facility, all testing is done and reviewed by the facility’s Quality Team which typically involves a radiochemist and radiopharmacist. For lyophilized drugs intended to be later mixed with a nuclide at time of use, the cold product is tested at the GMP manufacturing facility for identity, nonradioactive purity, and microbiological purity. After aseptic preparation in a USP-compliant radiopharmacy, a radiopharmacist assesses radiochemical and radionuclidic purity, releasing the product for patient administration.

While much attention is paid to sterility in this process, microbial contamination is a very unusual problem with radionuclide agents. First, all the components are typically sterile. The environment and procedure are sterile and highly monitored and controlled, aseptic technique of personnel is routinely assessed, and the agent is tested very shortly after production. All these factors mitigate against significant microbial contamination. The agent is used quickly after manufacture so that there is little time for bacterial growth. To some extent the radioactivity may also inhibit bacterial growth; that is partially why the FDA allows radiopharmaceuticals to be administered before full sterility results have returned. A far more common issue than sterility is the problem of an unsuccessful radiosynthesis such that insufficient purity is achieved, or contaminants are detected resulting in an insufficient dose. Supply or contamination problems with the radioisotope can contribute to failed syntheses. Once the cGMP synthesis has been worked out, the FDA requests that three consecutive qualifying runs be performed with documentation that the procedure produces a high-quality stable agent in a consistent manner as part of the Investigational New Drug (IND) application.

5.0. Toxicity Studies

An important part of the IND process is to conduct toxicity studies on the test agent. However, radiopharmaceuticals fall in a special category based on their extremely low mass dose. Typically, Phase 1 studies of most drugs require sufficient toxicity data in two species of animals before an IND will be issued. Based on the “microdose” used for most radiopharmaceuticals, the Phase 0 approach can be used in which only one species is employed in limited toxicity studies. Toxicity studies should be performed in the same manner as the agent is intended to be used in humans. Typically, the dose of radiopharmaceuticals is limited by radiation exposure, so that a maximum dose is easy to establish prior to Phase 1 testing based on dosimetry from prior animal biodistribution studies. Establishing the dose for a toxicity study is not an exact science and it is highly recommended that a pre-IND meeting with the FDA be held to obtain the agency’s opinion on the suitability of proposed toxicity studies.

Briefly, the proposed maximum dose (or several-fold that amount if chemically possible) will be injected and animals will be followed for up to 10 half-lifes of the agent under study. In the case of mice or rats, the animals will be euthanized at fixed time points after injection and studied for histologic evidence of toxicity. Serum markers of toxicity (complete blood counts, electrolytes, and enzymes) will be obtained, and basic clearance studies performed. For targeted agents, xenograft models can be used if the tumor expresses the human target. However, the most relevant model for toxicity for many agents targeted in humans are non-human primates (NHP), the closest biologic relative of humans. Typically, NHP studies may not be necessary for small molecule radiopharmaceuticals because the clearance is rapid, and the chemical dose is so low. Minimal single species toxicity is typically all that is required but the advice of the FDA should be sought in each individual case since the agency’s opinions can change over time. Prior to submitting an IND, developers can seek a “pre-IND” meeting in which the FDA informally addresses questions from the developer. These are usually limited to less than one hour and typically only one pre-IND meeting is granted per submitted agent.

6.0. The Clinical Protocol.

Earlier than one might expect, the Phase 1/2 initial clinical trial concept should be conceived and discussed. Investigators and their teams should have a good idea about the design of the clinical trial they wish to conduct to validate a molecular imaging agent, which is often referred to as an “activating clinical trial”. In the current use case, a targeted small molecule imaging agent is employed to detect a specific type of cancer. Although the major purpose of the initial first-in-human (FIH) study is to establish safety, a secondary endpoint is usually a validation that the agent is doing what it is supposed to be doing, namely detecting a particular type of tumor. Therefore, a component of the protocol is histologic validation of imaging findings, either by biopsy or surgery, of the tumor of interest. This is vital to establishing the sensitivity and specificity of the agent and will strongly influence the approved use labeling of the drug.

Trial size is somewhat contingent on how good the agent is expected to be. If, for instance, the disease entity if very hard to identify with current methods and the proposed agent is 70-80% effective, the number of patients needed to show superiority over conventional imaging may be small. If the current state of detection is relatively good, then a higher threshold will be needed to demonstrate the superiority of a new agent and it may require more patients to achieve statistical significance.

Despite promising results with in vitro and in vivo studies, the true performance of a radiolabeled imaging agent may differ significantly in humans due to complex interactions exclusive to people. Circulating proteins, immune-mediated responses and variations in cellular permeability are a few potential factors that could alter a radiotracer’s course in the body which may not be replicated in animal models [8]. Therefore, small pilot size numbers of 5-10 clinical subjects may be prudent in preliminary studies before investing in larger population trials. These types of clinical trials are called “activating clinical trials” because they are used in conjunction with the IND application to “activate” the IND. Follow on studies may be completely different in design but generally a small simple study is what is needed for activation.

6.1. The Investigational New Drug Application (IND)

Assuming that one is not following the RDRC pathway, the new agent will require an IND. An IND consists of any preclinical data obtained with the agent, any toxicity and clearance data and a section called Chemistry Manufacturing and Controls (CMC) in which details of the chemical manufacture of the agent under cGMP conditions are described. Three qualifying preparative runs need to be included in the CMC to demonstrate methodological reproducibility and the IND will include the activating Phase 1 clinical trial. Although the IND application sounds straightforward, it is very labor intensive and should be assembled by people experienced in putting together INDs. Since the document will be scrutinized by highly sophisticated FDA staff, it should be compiled according to the exact template and in the format suggested by the FDA, which currently involves a very specific electronic submission.

Accompanying the IND submission is the proposed activating clinical trial protocol. The FDA has up to 30 days to respond to an IND proposal and may ask for additional information or ask questions of the investigators for clarity. The back and forth between the investigator and the FDA can, therefore, extend well beyond the 30-day limit. Nonetheless, the process is generally in the interest of the safety of the patient and is important to the ultimate success of the agent.

6.2. Phase 1 and Phase 2 Testing

Although Phase 1 studies are generally used to establish dose limiting toxicity (DLT) for most drugs, that is not the case with imaging agents that are radiolabeled. All Phase 1 and 2 studies must be completed with cGMP compounds. Because of the nature of imaging agents, it is usually necessary to have an outside laboratory create GMP substrate that is then radiolabeled on site or in a local cGMP facility. The on-site cGMP facility is responsible for producing and releasing the agent for patients. Therefore, both a radiochemist and a radiopharmacist are critical members of the Phase 1/2 team.

In such studies, the dose of the agent is already fixed by the calculated radiation exposure limits to the organs. Typically, one organ (e.g., kidney) is exposed disproportionately to the others and this limits the maximal radiation dose exposure to the patient. That is, the initial dosing chosen is usually determined by preclinical biodistribution studies in which the radiation dosimetry for humans is extrapolated from mice studies. However, because the estimates from mice can be misleading, it is common practice to evaluate the first five human patients and re-calculate the estimated dosimetry from the PET scan data in these patients. Small adjustments in dose (both upward and downward) are then made for subsequent patients.

Once this initial testing is completed the goal of the Phase 1 study is to validate that the imaging agent is imaging its intended target in humans in concordance with the ultimate labeling indications of the agent. This is usually accomplished by scanning patients with the disease or cancer of interest and evaluating uptake with respect to target expression by immunohistochemistry. While not mandatory, imaging lesions that are not actually cancer, or are of a different cancer type to establish specificity is highly desirable. The phase I study is primarily a safety study evaluating side effects of the imaging agent. Each patient is carefully monitored with vital signs and appropriate blood work before and after administration. Since our use case is a small, targeted molecule the adverse events are presumed to occur within days of the injection. Acute monitoring is therefore undertaken for patient safety. For longer-lived agents (e.g., radiolabeled antibodies), the period of safety monitoring will correspond to the serum clearance of the agent. Typically, evaluations are also performed later (often by phone) with the patient, within 2-4 weeks post injection, to evaluate for any unexpected, delayed reactions. Deaths due to disease or other causes may occur within this time period and it is important to ascertain whether the test drug had any impact on this severe adverse event, a process known as attribution. Phase 1 studies for safety are typically small numbering less than 25 patients and often only 10 or so. A Phase 1 or 2 trial may include a “stopping rule” that ends the study if unexpected early toxicity is observed.

Unless the FDA has previously approved a Phase 2 study, permission must be sought to expand the patient population in a Phase 2 study. In order to streamline this process, once safety is shown, Phase 1/2 studies are often written and approved by the FDA from the outset so that the agent can continue to be tested in a larger cohort of patients (n=30-50) without interruption, and with the Phase 2 focus directed less on the safety aspect and more on the efficacy of the agent.

6.3. Phase 3 and Approval

Academic centers can seek NDA approval of an imaging agent. In a recent example, ⁶⁸Ga-PSMA-11 was approved after submission by UCSF and UCLA. Another example is the ⁶⁸Ga-DOTATOC from the University of Iowa. However, most agents require the assistance of a company to make a viable agent. A counter example to ⁶⁸Ga-PSMA-11 is ¹⁸F-DCFPyL, another PSMA targeted agent in which Phase 1 and 2 studies were completed in academic centers and then a company (Progenics, which was then sold to Lantheus) conducted the Phase 3 study which resulted in its approval in May 2021, allowing the agent to reach the market (Figures 11-13). Approval must be sought from the FDA to proceed to Phase 3 after review of the Phase 1 and 2 data. Occasionally, if the data is compelling the agent can be “fast tracked” by the FDA. Phase 3 clinical trials are multi-institutional and involve a relatively large number of patients. They are sometimes referred to as pivotal trials because they include enough patients to capture severe side effects as well as a diversity of conditions and are therefore, pivotal to the decision to approve the agent. Few academic sites have the resources to take the agent this far. Moreover, production must be uniform across sites and this mandates sophisticated production logistics with a network of cGMP labs across the country, generally outside the capabilities of academic centers. Therefore, most agents require a corporate partner to progress to an NDA. However, in some cases the existing literature is sufficient for it to be submitted as evidence of efficacy and safety.

Figure 11.

¹⁸F-DCFPyl scan in a patient with recurrent prostate cancer after radiation therapy. Uptake is seen in several pelvic lymph nodes (arrows) for which the patient was treated with radiation therapy to the pelvis. Note the prominent uptake in the normal salivary and lacrimal glands as well as the kidneys an bladder which are the routes of excretion for this agent.

Figure 13.

¹⁸F-DCFPyl scan in a patient with widespread metastatic disease and local recurrence in the prostate.

By this time in the history of the imaging agent there is usually a large amount of enthusiasm for the agent in the medical community. Funding and investing are relatively easier at this point. Moreover, participating in such Phase 3 trials can be profitable for medical institutions and therefore, accrual is not usually a problem. However, the FDA mandates careful auditing of all results and this necessitates a large number of employees or contract research organizations (CROs) to handle the data. Even after accrual is over and follow up has been completed compiling the data for the FDA is a time consuming and expensive process and often takes over a year. In order to process the application, the company must also pay a significant sum of money to the FDA to comply with the Prescription Drug User Fee Act (PDUFA). All of these factors favor medium to large size companies with extensive capital resources and experience dealing with the FDA. The FDA will consider the data submitted from Phase 1, 2 and 3 studies and ask for any additional data. They may ask for blinded readings of PET imaging results by neutral radiologists to demonstrate utility of the study agent outside of tertiary care health care environments. Such readings may be required to be performed either with or without patient historical file data listing other tests and observations that have been done. If safety and efficacy data is sufficient, the FDA will then issue a New Drug Application (NDA) indicating approval of the drug for sale to the public.

6.4. A continuing process.

One might be forgiven for thinking that the NDA is the end of the process, but, in fact, the most important aspects are still to come. Once a company has received its NDA, it applies for (and usually receives) permission to charge a “pass through” fee to patients undergoing the examination. The patient can be billed for the test and depending on their insurance it may be reimbursed in part or in full. However, very large payors such as Medicare and Medicaid reserve judgement on reimbursement and the drug is not fully viable until it receives approval for reimbursement from these agencies. The period in which “pass through” payments occur is time limited. During that period application to the Center for Medicare and Medicaid Services (CMS) is made to determine reimbursement. Many insurance companies follow the lead of CMS and therefore their approval at this stage is critical for the economic viability of the new product. CMS is interested in medical efficacy but is also interested in cost efficacy. For instance, they may ask companies what other tests, currently in use, may be phased out if this test is implemented. This end-use aspect is extremely important to consider much earlier in the process because if the imaging agent is simply viewed as an “add on” cost to existing diagnostic tests, it may face difficulty at this stage at the CMS. Again, a demonstrable improvement in patient management is probably the key outcome to enable approval, general application and reimbursement for the new agent. However, other outcomes such as improved quality of life or faster access to treatment can also be considered desirable endpoints.

6.5. Medical Advances and Other Threats

Assuming the agent receives CMS approval and is being used in the medical community, a new agent can face many unforeseen threats. For instance, any occurrence of rare but well publicized adverse events related to the agent can dim enthusiasm for its use even after several million doses and for particularly spectacular situations can cause the agent to be withdrawn from the market to avoid liability claims. Medical science also continues to advance. The process of approval can take many years, during which time, better or different agents may come along as competitors within an already small market. Moreover, improvements in imaging technology can render the need for any specific agent superfluous. Ideas surrounding the disease itself can change, particularly if a new diagnostic test or therapy is introduced. This can make it difficult to make a profit and if sales lag, companies, whose first duty is to their stockholders and investors, may be forced to take their agent off the market.

Hopefully, however, the new molecular imaging agent will become a success and will be widely used for many years. Moreover, it may be incorporated into practice guidelines, which mandate its use into the future. Naturally, the patent, which is granted for 17 years, may expire and new competitors with generic versions may enter the market, although radiopharmaceuticals are complex enough that few companies dare to enter the field and so a very successful agent may enjoy decades of economic success and become a cornerstone of diagnosis and therapy for patients. However, “knock-offs” are to be expected for particularly successful agents.

7.0. Conclusion

The purpose of this review was to outline the path from discovery to approval of a targeted molecular imaging agent. While “clinical translation” is a simple concept, the reality is far from simple and requires persistence, diligence, focus and financial resources. Very few agents make it through the entire gauntlet outlined here and such a gauntlet may also vary considerably with the nature of the agent. It is rare to find individuals who have taken a product from inception to approval. Rather, most of us become experts in our segment of the process. Although the process is daunting and may seem discouraging, there are considerable potential benefits to new imaging agents, and these are likely to dramatically increase in this age of molecular medicine. At the most fundamental level, novel imaging agents can help clinicians visualize disease that is otherwise cryptic, allowing for earlier interventions, which may improve outcomes. This is underscored by the marked improvement in diagnosis, staging, surveillance, and treatment response with the PSMA-targeted imaging agents for men with prostate cancer (Figure 11,12). Radiation oncologists can include or deliver higher radiation doses to disease that would have otherwise been excluded from conventional treatment volumes. An added benefit to these novel agents is that they can be engineered to deliver therapeutic radioisotopes. Not all theranostic partners will be successful, but the overall survival and quality of life improvements observed with [¹⁷⁷Lu]-DOTATATE [95] and [¹⁷⁷Lu]-PSMA-617 [96] speaks to the magnitude of potential impact to humanity, and certainly well worth the considerable efforts that are required.

Figure 12.

¹⁸F-DCFPyl scan in a patient with recurrent prostate cancer after surgery. The pointed shape of the bladder is due to prior resection of the prostate. Although multiple “hot spots” are seen only one (arrow and circle) represents recurrent cancer while the other areas are due to radioactivity in the ureters.